Cursor AI has disclosed a security breach involving a modified version of the Claude Code VS Code extension. The malicious package, version 2.1.313, was distributed through Cursor's marketplace and affected up to 5,000 users during an approximately 11-hour window on October 29, 2026. The compromised extension attempted to install a persistent SSH backdoor and stole sensitive data including public IP addresses, hostnames, and environment variable names. Cursor shut down its marketplace to remove the harmful package, though Anthropic has not yet publicly commented on the incident.
The incident underscores growing security concerns around AI coding tools that are rapidly becoming standard in developer workflows. Security researchers confirmed the tampered extension was distributed under Anthropic's legitimate extension ID. Users are now advised to remove any unauthorized SSH keys and rotate their credentials. The hackers' identities remain unknown, and the event highlights the risks associated with relying on third-party extensions in software development environments.
Separately, Anthropic has flagged its own AI model as a significant risk factor ahead of its planned IPO. In its S-1 filing, the company warned that Claude could become misaligned and cause catastrophic harm. Anthropic uses Claude for internal work and acknowledged it may lose visibility into what the model does. Investors are being urged to take these warnings seriously as the company moves toward going public.
Amazon's third-quarter results revealed the financial impact of its AI investments. Net income dropped to $20.7 billion from $21.2 billion a year earlier, partly because the prior year included $9.5 billion in pre-tax gains from its stake in Anthropic. AWS revenue grew 39.3% year-over-year, its fastest pace in nearly five years, while Amazon Ads revenue rose 12% to $19.6 billion. Total revenue reached $201.9 billion, beating forecasts, but free cash flow turned negative at $42.4 billion over the past 12 months, prompting a 7% drop in after-hours trading.
Asian markets also felt the effects of AI-related concerns. South Korea's KOSPI fell 3.48% to 5,730.74 after weak earnings from US tech giants raised doubts about AI investment returns. Samsung Electronics lost 3.02% and SK hynix dropped 7.08%. Samsung warned that its high-bandwidth memory supply for 2027 is already sold out, with tight conditions potentially lasting through 2028. Together, Samsung and SK hynix control roughly two-thirds of the global memory market.
On the education and training front, Scale and UMSL launched a free 12-hour self-paced AI training course for US teachers. The program includes applied work where participants create an artifact tied to a specific learning objective and earn a certificate upon completion. Meanwhile, a new framework called ANCHOR shows promise in improving location prediction using large language models by first estimating category-level intent before identifying specific destinations, outperforming traditional methods on real-world datasets.
Key Takeaways
- <ul><li>Cursor AI shut down its marketplace after a modified Claude Code VS Code extension (version 2.1.313) compromised up to 5,000 users with an SSH backdoor over roughly 11 hours on October 29, 2026.</li><li>Anthropic has not publicly commented on the extension breach or warned customers about the incident.</li><li>Anthropic's S-1 filing warns that its Claude AI model could become misaligned and cause catastrophic harm, flagging model alignment as a major stock risk ahead of its IPO.</li><li>Amazon's Q3 net income fell to $20.7 billion from $21.2 billion, missing expectations due to the absence of $9.5 billion in Anthropic-related gains from the prior year.</li><li>AWS grew 39.3% year-over-year in Q3, its fastest rate in nearly five years, while Amazon Ads revenue rose 12% to $19.6 billion.</li><li>Amazon's free cash flow turned negative at $42.4 billion over the past 12 months due to increased AI investment, causing a 7% drop in after-hours stock trading.</li><li>South Korea's KOSPI fell 3.48% to 5,730.74 as weak US tech earnings raised concerns about AI investment returns, with SK hynix dropping 7.08% and Samsung losing 3.02%.</li><li>Samsung warned that high-bandwidth memory supply for 2027 is already sold out, with tight conditions potentially persisting through 2028, raising risks of higher server and PC costs.</li><li>Scale and UMSL are offering a free 12-hour self-paced AI training course for US teachers, including applied work and a completion certificate.</li><li>European police forces in France, Italy, and Spain are purchasing AI video analysis tools to track rioters and manage public disorder during protests.</li></ul>
Cursor AI reports security breach in Claude Code extension
Cursor AI identified a security incident involving a modified Claude Code VS Code extension. The malicious version 2.1.313 was distributed through Cursor's marketplace and affected at most 5,000 users. The compromised package contained a malicious loader that attempted to install a persistent SSH backdoor. Cursor shut down its marketplace to remove the harmful extension.
Claude Code VS Code extension compromised with SSH backdoor
Cybersecurity researchers found a modified Claude Code VS Code extension on Cursor's marketplace that installed a persistent SSH backdoor. The malicious version 2.1.313 was distributed under Anthropic's legitimate extension ID and affected up to 5,000 users between 2:27 a.m. and 1:35 p.m. PT on October 29, 2026. The extension stole sensitive data including public IP, hostname, and environment variable names. Users are advised to remove the unauthorized SSH key and rotate their credentials.
Hackers hijack Claude Code extension in new AI security risk
Hackers hijacked a Claude Code extension available on Cursor's VS Code marketplace for about 11 hours. The tampered extension collected user computer information and attempted to install a persistent backdoor using SSH. Security experts say the incident highlights growing risks as AI coding tools become standard in developer workflows. The hackers' identities remain unknown.
Anthropic VS Code extension backdoored on Cursor for 12 hours
An official Anthropic Claude model extension available on Cursor was backdoored for nearly 12 hours, threatening up to 5,000 developers' machines. The malicious version 2.1.313 was distributed through Cursor's marketplace. Anthropic has not yet commented on the incident or warned customers. The event highlights risks of using third-party extensions in software development.
KOSPI drops 3.48% amid US tech earnings and AI concerns
South Korea's KOSPI fell 3.48 percent on Friday, closing at 5,730.74, down 206.92 points. The decline was driven by weak earnings from US tech giants like Alphabet and Amazon, which raised doubts about AI investment returns. Samsung Electronics lost 3.02 percent and SK hynix dropped 7.08 percent. The Kosdaq index also fell 3.83 percent to 724.38.
SK Hynix and Samsung stocks fall as AI sell-off hits Asian markets
South Korean stocks fell on Friday after weak U.S. tech earnings raised concerns about AI investments. The Korea Exchange activated a rare sell-side trading halt for the KOSPI index. SK Hynix and Samsung Electronics shares tumbled. The Nikkei 225 also fell 2.12 percent. The sell-off spread beyond South Korea into other Asian markets.
ANCHOR framework improves next location prediction using language models
Researchers developed ANCHOR to predict a user's next destination from check-in data. The framework uses large language models but first estimates category-level intent instead of guessing specific places. It combines category anchors, visit history, and user preferences to find the right location. Tests on three real-world datasets showed ANCHOR outperformed traditional and LLM-based methods.
European police adopt AI video tools to track rioters
Police forces in France, Italy, and Spain are buying artificial intelligence tools to analyze video feeds during protests. The tools are designed to tackle a new kind of threat emerging on social media. The agencies are scrambling to acquire these capabilities for managing riots and public disorder.
Samsung warns AI memory chip shortage could worsen in 2027
Samsung said its high-bandwidth memory supply for 2027 is already sold out. The company reported record third-quarter operating profit of 107.4 trillion won. Tight memory conditions could last through 2028 and may raise server and PC costs. Samsung and SK hynix control about two-thirds of the global memory market.
Anthropic warns its own AI model misalignment is biggest stock risk
Anthropic filed its S-1 form ahead of a planned IPO. The company said its AI model could become misaligned and cause catastrophic harm to people. Anthropic uses its own Claude AI for internal work and warned it may lose visibility into what Claude does. Investors are urged to take the company's risk warnings seriously.
Amazon's AI Investments Impact Income as AWS and Ads Grow
Amazon's third-quarter earnings missed expectations with net income dropping to $20.7 billion from $21.2 billion last year. The company lacked the $9.5 billion in pre-tax gains from AI startup Anthropic that boosted last year's results. AWS grew 39.3% year-over-year, its fastest rate in nearly five years, while Amazon Ads revenue rose 12% to $19.6 billion. Amazon stock dropped 7% in after-hours trading as investors expressed concern over negative cash flow of $42.4 billion over the past 12 months.
Amazon Q3 Revenue Beats Forecasts as Operating Profit Surges
Amazon reported third-quarter revenues of $201.9 billion, up 12% year-on-year, exceeding its forecast range. Operating profit jumped to $28.1 billion from $17.4 billion in the same period in 2025, surpassing the company's guidance. The strong performance was driven by growth at Amazon Web Services and the North American retail business. However, free cash flow turned negative due to increased AI investment.
AI Brings Both Risks and Opportunities for Women in Insurance
Artificial intelligence poses particular challenges for women in insurance because they often hold administrative roles susceptible to automation. However, AI also creates opportunities for women to move into strategic roles and develop new skills. Women in insurance are more likely to be in roles at risk of automation but also in roles less likely to be automated, such as data analysis and customer service. The report recommends women focus on critical thinking and problem-solving skills while AI is described as a tool to aid decision-making rather than replace human judgment.
Scale and UMSL Offer Free AI Training for US Teachers
Scale and UMSL announced a free 12-hour self-paced AI training course for US teachers. The course includes applied work where participants create an artifact for a specific learning objective. Completing the course earns participants a certificate and digital badge. The program aims to help teachers develop AI skills, though the listed section dates predate the announcement.
Sources
- Cursor AI Says Co Identified Security Incident Affecting A Modified Claude Code Versus Code Extension Impacting At Most 5,000 Users
- Official Claude Code VS Code Extension Compromised to Install Persistent SSH Backdoor
- Hackers hijacked a Claude Code extension, exposing yet another new AI security risk
- Anthropic VS code extension on Cursor backdoored for nearly 12 hours
- KOSPI falls 3.48% as US tech earnings fuel AI investment concerns
- SK Hynix, Samsung Tumble in South Korea as Wall Street's AI Sell-Off Rattles Asian Markets
- Category-Anchored Large Language Models for Next POI Recommendation
- France/Italy/Spain • European police use AI video-tracking tools to tackle rioters
- Samsung Warns AI Memory Shortage Could Worsen in 2027
- The Biggest Risk to Anthropic Stock Is the Company Itself
- Unrealized AI Investments Drag Amazon’s Income, But AWS And Ads Are Still Growing
- Amazon beats Q3 outlook with operating profit up 61%, FCF turns negative on AI investment
- AI presents risks, opportunities for women
- Scale and UMSL announce free U.S. teacher AI training
Comments
Please log in to post a comment.